Concurrent-distributed programming techniques for SAT using DPLL-stålmarck
نویسندگان
چکیده
This paper reports our work on application of concurrent/distributed techniques to SAT. These were investigated using each of the following methods: the DPLL [5] algorithm, the dilemma rule of the Stålmarck’s algorithm [12] and a concurrent/distributed hybrid SAT solver: DPLL-Stålmarck, using a combination of the DPLL and the dilemma rule based algorithm. The prototypes have been implemented using Alice [10], an SML based language with support for distribution and concurrency. Our prototype framework allows for rapid-prototyping of and experimentation with application of various concurrent/distributed programming techniques to SAT. The emphasis is not on building an industry-standard SAT solver, but rather an investigation of the efficacy of use of these techniques for SAT at various levels of granularity. We discuss in detail the workings of DPLL-Stålmarck. DPLL and Stålmarck are complementary methods given their depth-first and breadth-first approaches. A single threaded execution limits the possibilities of combining the two methods. We have used our prototype framework to explore the use of concurrent/distributed programming techniques for achieving cooperation between the two methods. Empirical evaluation is in progress using the metrics of time, size of search space, and number of machines. The work reported is part of an ongoing project investigating the use of concurrent-distributed programming techniques for theorem proving
منابع مشابه
Parallelizing DPLL in Haskell
We report on a case study of implementing parallel variants of the DavisPutnam-Logemann-Loveland algorithm for solving the SAT problem of propositional formulas in the functional programming language Haskell. We explore several state of the art programming techniques for parallel and concurrent programming in Haskell and provide the corresponding implementations. Based on our experimental resul...
متن کاملActin - Technical Report
The Boolean satisfiability problem (SAT) can be solved efficiently with variants of the DPLL algorithm. For industrial SAT problems, DPLL with conflict analysis dependent dynamic decision heuristics has proved to be particularly efficient, e.g. in Chaff. In this work, algorithms that initialize the variable activity values in the solver MiniSAT v1.14 by analyzing the CNF are evolved using genet...
متن کاملBuilding State-of-the-Art SAT Solvers
The area of Propositional Satisfiability (SAT) has been the subject of intensive research in recent years, with significant theoretical and practical contributions. From a practical perspective, a large number of very effective SAT solvers have recently been proposed, most of which based on improvements made to the original Davis-Putnam-Logemann-Loveland (DPLL) backtrack search SAT algorithm. T...
متن کاملOptimizing the Initialization of Dynamic Decision Heuristics in DPLL SAT Solvers Using Genetic Programming
متن کامل
Satisfiability Checking of Non-clausal Formulas Using General Matings
Most state-of-the-art SAT solvers are based on DPLL search and require the input formula to be in clausal form (cnf). However, typical formulas that arise in practice are non-clausal. We present a new non-clausal SAT-solver based on General Matings instead of DPLL search. Our technique is able to handle non-clausal formulas involving ∧,∨,¬ operators without destroying their structure or introdu...
متن کامل